We investigate the influence of wildfire smoke aerosols on cloud microphysics and precipitation using a coupled aerosol-cloud microphysics-meteorology model WRF-Chem-SMOKE. The Wildfire Automated Biomass Burning Algorithm products are used to compute “online” hourly size- and composition-resolved smoke emission fluxes during Canadian boreal wildfires in the summer of 2007. Comparisons with Moderate Resolution Imaging Spectroradiometer aerosol optical depth, Ozone Monitoring Instrument aerosol index, and Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation vertical feature mask demonstrate that WRF-Chem-SMOKE captures both the horizontal and vertical spatial distribution of smoke. However, estimated smoke emissions result in much lower aerosol optical depth values than those of observations (by about tenfold). Modeling experiments with varying amounts of smoke emissions of 5 to 10 times as high as the original load reveal that low smoke load favors the collision-coalescence process at a certain stage, leading to either positive or negative changes in the cloud water path (CWP) relative to smoke-free conditions. For high smoke emissions, changes in CWP are positive, as large as 0.5 kg/m2. A domain-integrated increase in CWP is proportional to smoke loading. By contrast, both positive and negative changes in the rain water path (RWP) and the snow water path (SWP) are found. While domain-integrated changes in RWP are negative, those in SWP go from negative to positive under a high smoke load. Higher smoke loadings suppress precipitation initially, because of smoke-induced reduction of the collision-coalescence and riming processes, but ultimately cause an invigoration of precipitation. We found that precipitation is highly sensitive to 3-D smoke fields and varies in a nonlinear manner with smoke loads.